Principal component analyses were applied to carcass records of 367 animals of a F2 swine population. Starting from 33 traits, 17 showed variation lower than 0.7 (eigenvalue lower than 0.7) and significant simple linear correlations with the other traits, indicating they were redundant. These results suggest the following variables to be used in future carcass studies: slaughter age; right side carcass weight; carcass length by the Brazilian Method of Carcass Classification; higher backfat thickness at boston shoulder area and in the midline; loin depth; spleen, heart, cold right side carcass, ham, boston shoulder, picnic shoulder, loin (bone-in), spareribs, jowl, sirloin and kidney weights.
carcass; correlations; discard of variable; multivariate analysis
